Thermal Sensor Fusion for Real-Time Arc Weld Quality Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing welding technologies lack real-time, non-destructive methods for quality inspection during the welding process, leading to increased cycle time, labor, and potential safety risks due to off-line quality inspection and destructive testing.
Innovation Solution
Implementing a thermal imaging-based sensor fusion system with AI/ML models to predict welding quality in real-time by correlating sensor data with final quality data, allowing for immediate adjustments and continuous model improvement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If off-line quality inspection and destructive testing are used, then quality assessment is thorough, but production cycle time increases and labor requirements increase
Solution Approach 1:
The patent replaces mechanical destructive testing and off-line inspection methods with optical sensing (thermal imaging cameras) and computational analysis (sensor fusion algorithms). This substitution enables non-contact, real-time quality assessment without physical intervention, thereby eliminating the time loss associated with destructive testing while maintaining measurement precision through multi-sensor data integration.
Solution Approach 2:
The system performs preliminary quality assessment during the welding process itself by continuously monitoring thermal signatures and other process parameters. By detecting quality issues in real-time rather than after completion, the system enables immediate corrective actions, preventing defective welds from being produced and eliminating post-process inspection time.
2Reliability
If destructive testing is performed, then quality verification is comprehensive, but additional lead time is added due to time-consuming testing
Solution Approach 1:
The patent replaces time-consuming destructive testing with optical and thermal sensing systems that capture weld quality information in real-time. The sensor fusion algorithm processes thermal imaging data, visible light images, and process parameters to verify weld quality without physical contact or additional time, achieving both comprehensive verification and zero lead time addition.
Solution Approach 2:
The system creates a digital replica of the weld quality characteristics through thermal imaging and sensor data fusion, eliminating the need for physical destructive testing. This digital copying approach provides comprehensive quality verification by analyzing thermal signatures and process parameters that correlate with weld integrity, all without adding lead time.
3Loss of time
If real-time quality monitoring is implemented, then immediate quality feedback is provided, but system complexity increases
Solution Approach 1:
The patent merges multiple sensing modalities (thermal imaging, visible light imaging, process parameter sensors) into a unified monitoring system. By combining these sensors and their data streams, the system achieves comprehensive real-time quality feedback while managing complexity through integrated data processing algorithms that fuse multiple data sources into actionable quality information.
Solution Approach 2:
The sensor fusion algorithm acts as an intermediary that processes and integrates data from multiple complex sensors, transforming raw thermal and optical signals into meaningful quality metrics. This intermediary processing layer simplifies the overall system by providing a unified interface between diverse sensors and the control system, enabling real-time feedback without proportionally increasing complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables real-time, non-destructive quality judgment with high accuracy and precision, reducing production interruptions and enhancing welding quality through continuous model adaptation.
Implementation Method 1
a thermal imaging camera to capture thermal signatures during the welding process
Data Source
AI summary
Systems and methods described herein involve intaking sensor data associated with an arc weld from a robotic welding process, the sensor data involving thermal imaging data; executing a machine learning model on the sensor data, the machine learning model configured to output predicted internal parameters of a weld seam of the arc weld, predicted surface parameters of the weld seam of the arc weld, and a predicted quality of the arc weld associated with the predicted internal parameters and the predicted surface parameters; and modifying parameters of the robotic welding process of the arc weld based on the output predicted quality, the output predicted internal parameters, and the predicted surface parameters.